CN103889053A - Automatic establishing method of self-growing-type fingerprint - Google Patents

Abstract

The invention discloses an automatic establishing method of a self-growing-type fingerprint, and relates to position fingerprint locating technologies. The automatic establishing method aims to solve the problem that a large amount of workload is needed in fingerprint establishing and maintaining in the fingerprint locating process. The offline part of the method comprises the steps that a seed zone is established, a fingerprint of a current service area is determined, RSS estimation is conducted on corresponding reference points according to user position prediction results and RSS values uploaded by the online part, a fingerprint is updated, and after the fingerprint grows to a certain degree, the fingerprint of the current service area is determined again. The online part of the method comprises the steps that a receiver at a user-side collects RSS vectors; a user conducts fingerprint locating by using the fingerprints provided by the offline part; the position of the user is predicted by using the filtering algorithm; whether the user leaves the service area or not is judged. By means of the automatic establishing method, the workload needing to be consumed in a fingerprint locating system in the fingerprint establishing process is greatly reduced, the establishing speed of the fingerprint is increased and is beneficial for popularization and application of the fingerprint locating technologies in the commercialization aspect.

Description

A kind of autonomous method for building up of fingerprint image from growth formula

Technical field

The present invention relates to a kind of fingerprint image method for building up, relate to Indoor Position Techniques Based on Location Fingerprint.

Background technology

The enforcement of location fingerprint location generally can be divided into two stages: the first stage is training/off-line phase, groundwork is the signal characteristic parameter that gathers the each reference node of required locating area position, such as signal strength, multipath phase angle component power etc., form location fingerprint database by a corresponding one group of finger print information specific position.Second stage is location/on-line stage, utilizes receiver to measure and receives the parameter of signal, adopt matching algorithm determine with database in which is organized data and matches, thereby draw user's physical location.In the time adopting location fingerprint localization method, classical positioning experiment flow chart as shown in Figure 2: wherein, (RP ₁, RP ₂..., RP _n) represent the 1st to n reference point, RSSi _j(i=1 ..., n; J=1 ..., T) and j RSS signal phasor measuring at i reference point place of representative.(TP ₁, TP ₂..., TP _m) represent the 1st to m reference point.In positioning stage, WLAN navigation system is taking Radio Map as basis, and the wlan client being positioned at needs carries out the real-time sampling of spacing wave, and utilizes mobile computing environment and data transmission environments transmission and the calculating sampling data of WLAN.Computational process is mainly carried out search and the location of locus by applying the search of specific signal space and matching algorithm, draw the position prediction result to sampled data, completes the location of locus.Under actual conditions, building has different physical dimensions and internal structure, and in order to simplify the task of training stage, the selection of reference point often needs suitably to select according to building structure.Meet under equally distributed condition reference point is approximate, the effect of distance of adjacent reference point positioning precision, and along with the increasing of reference point, positioning precision is higher.In fingerprint location process, the foundation of fingerprint image and maintenance process need to drop into a large amount of workloads, and this is fingerprint location technology main technical barrier aspect commercialization.

Summary of the invention

The object of this invention is to provide a kind of autonomous method for building up of the fingerprint image of growth formula certainly based on user's uploading data, need to drop into the problem of a large amount of workloads to solve foundation and the maintenance process of fingerprint image in fingerprint location process, thereby affect the application of fingerprint location technology.

The present invention solves the problems of the technologies described above the technical scheme of taking to be:

From the autonomous method for building up of fingerprint image of growth formula, the described autonomous method for building up of fingerprint image from growth formula is realized according to the following steps:

Step 1, off-line part

Step 1 (one), foundation " seed zone ":

Measure a basic fingerprint image at the key position of building, be referred to as " seed zone "; Wherein, the user in described seed zone can carry out fingerprint location;

Step 1 (two), determine current service district fingerprint image:

The time real growth district extending that current service district fingerprint image comprises " seed zone " and the reference point that calculates according to online certain customers end uploading data forms; Described uploading data refers to customer location coordinate and the field intensity value corresponding with this position, i.e. RSS vector; And current service district fingerprint image is offered to the user side of online part;

The detailed process of step 1 (three), fingerprint image growth:

The customer location of uploading up according to online part predicts the outcome and RSS value, and corresponding reference point RP is carried out to RSS estimation, upgrades fingerprint image, and fingerprint image growth, after fingerprint image grows into a certain degree, jumps into step 2 and redefine current service district fingerprint image;

Described customer location is uploaded information and is shown below:

A=(RSS _a1,RSS _a2,RSS _a3,…,RSS _aP,X _a,Y _a)

RSS in formula _axthe RSS value of x the AP that representative is uploaded, X _a, Y _abe respectively the coordinate information of the receipts machine estimated position reporting; The span of x is 1～P, and P is the total number of AP in localizing environment;

It is described that corresponding reference point (RP) is carried out to the process of RSS estimation is as follows:

To a large amount of lastest imformation A that upload, try to achieve reference point locations to be estimated (Xi, Y _j) with all estimated positions (Xa, Ya) between two-dimensional distance D _dis_ a, the line item of going forward side by side;

$D_{\mathrm{dis}\_}a=\sqrt{{(X_a-X_i)}^2+{(Y_a-Y_j)}^2}$

Then system is that each reference point to be recovered is found out a nearest with it K estimated distance: D _dis_ a ₁, D _dis_ a ₂, D _dis_ a ₃..., D _dis_ a _k, and by corresponding RSS vector in lastest imformation being averaged to obtain the RSS vector of reference point;

${{\mathrm{RSS}}_{\mathrm{ijp}}}^{\′}=\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{\mathrm{RSS}}_{a_{k}p}/K---\left(1\right)$

In formula, p represents No. AP, RSS _ijp' representing the estimated value of the RSS value of (i, j) number p AP of reference point, the value of p is between 1～P; By as shown in the formula weighing the maturity Tr that is resumed reference point:

$\mathrm{Tr}=\frac{\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{D}_{\mathrm{dis}\_}{a}_i}{K}---\left(2\right)$

Work as Tr<Tr _thtime, think that reference point is enough ripe, can add in fingerprint image and use; Wherein, Tr _thfor with reference to thresholding;

Step 2, online part

The receiver of step 2 (), user side gathers RSS vector;

The fingerprint image that step 2 (two), user use the step 1 (two) of off-line part to provide carries out fingerprint location;

Step 2 (three), utilize filtering algorithm predictive user position;

Step 2 (four), judge whether user leaves service area; Described service area refers to the region that current time fingerprint image covers;

Step 2 (five) if, upload RSS and current location, carry out fingerprint image growth operation for the step 1 (three) of off-line part; Otherwise, return to execution step two (), until user leaves the described service area of step 2 (four).

The key position of building described in step 1 refers to the region that the stream of people is larger or concentrated.

Described in step 1 (three), when growing into, fingerprint image to a certain degree refers to that maturity function Tr is less than its threshold T r _thtime current fingerprint image.

The invention has the beneficial effects as follows:

The use of the inventive method greatly reduces the workload that in fingerprint location system, fingerprint image process of establishing need to expend.The inventive method provides initial positioning service by the fingerprint image among a small circle set up in advance for user, binding site prediction algorithm, the renewal data upload that user is collected is to server, these type of data of Server Consolidation, time the real coverage that expands fingerprint image, without setting up large-scale fingerprint image in early stage, the inventive method has improved the construction speed of fingerprint image, is conducive to fingerprint location technology applying aspect commercialization.

Brief description of the drawings

Fig. 1 is fingerprint image based on the inventive method workflow block diagram from growing system; Fig. 2 is WLAN fingerprint location system flow schematic diagram in prior art; Fig. 3 is the system data storage node composition of realizing the inventive method, Fig. 4 be working-flow figure (register group store be K and upgrade the RSS vector of some data and upgrade a distance B apart from this reference point _dis_ a _x); Fig. 5 is the emulation experiment environment map that utilizes the inventive method; The simulation experiment result figure of Fig. 6 the inventive method, in Fig. 6:

Fig. 6 a shows 5000 users fingerprint image recovery after this scene, and in figure, abscissa represents the X-axis coordinate of experimental situation, and in figure, ordinate represents the Y-axis coordinate of experimental situation, and unit is rice;

Fig. 6 b shows 10000 users fingerprint image recovery after this scene, and in figure, abscissa represents the X-axis coordinate of experimental situation, and in figure, ordinate represents the Y-axis coordinate of experimental situation, and unit is rice;

Fig. 6 c shows 20000 users fingerprint image recovery after this scene, and in figure, abscissa represents the X-axis coordinate of experimental situation, and in figure, ordinate represents the Y-axis coordinate of experimental situation, and unit is rice;

Fig. 6 d shows 50000 users fingerprint image recovery after this scene, and in figure, abscissa represents the X-axis coordinate of experimental situation, and in figure, ordinate represents the Y-axis coordinate of experimental situation, and unit is rice;

Fig. 6 e shows 100000 users fingerprint image recovery after this scene, and in figure, abscissa represents the X-axis coordinate of experimental situation, and in figure, ordinate represents the Y-axis coordinate of experimental situation, and unit is rice;

Fig. 6 f shows 200000 users fingerprint image recovery after this scene, and in figure, abscissa represents the X-axis coordinate of experimental situation, and in figure, ordinate represents the Y-axis coordinate of experimental situation, and unit is rice.

Embodiment:

Embodiment one: as shown in Fig. 1,3 and 4, a kind of described in present embodiment realizes according to the following steps from the autonomous method for building up of fingerprint image of growth formula:

Step 1, off-line part

Step 1 (one), foundation " seed zone ":

Measure a basic fingerprint image (initial fingerprint figure) at the key position of building, be referred to as " seed zone "; Wherein, the user in described seed zone can carry out fingerprint location;

Step 1 (two), determine current service district fingerprint image:

The time real growth district extending that current service district fingerprint image comprises " seed zone " and the reference point (certainly grow and extend reference point) that calculates according to online certain customers end uploading data forms; Described uploading data refers to customer location coordinate and the field intensity value (RSS vector) corresponding with this position; And current service district fingerprint image is offered to the user side of online part;

Identical with seed zone when current service district starts, along with fingerprint image increases, positioning service district can expand gradually afterwards;

The detailed process of step 1 (three), fingerprint image growth:

The customer location of uploading up according to online part predicts the outcome and RSS value, corresponding reference point (RP) is carried out to RSS estimation, upgrade fingerprint image, fingerprint image growth, after fingerprint image grows into a certain degree, jump into step 2 and redefine current service district fingerprint image;

Described customer location is uploaded information and is shown below:

A=(RSS _a1,RSS _a2,RSS _a3,…,RSS _aP,X _a,Y _a)

RSS in formula _axthe RSS value of x the AP that representative is uploaded, X _a, Y _abe respectively the coordinate information of the receipts machine estimated position reporting; The span of x is 1～P, and P is the total number of AP in localizing environment;

It is described that corresponding reference point (RP) is carried out to the process of RSS estimation is as follows:

To a large amount of lastest imformation A that upload, try to achieve reference point locations to be estimated (Xi, Y _j) with all estimated positions (Xa, Ya) between two-dimensional distance D _dis_ a, the line item of going forward side by side;

$D_{\mathrm{dis}\_}a=\sqrt{{(X_a-X_i)}^2+{(Y_a-Y_j)}^2}$

Then system is that each reference point to be recovered is found out a nearest with it K estimated distance: D _dis_ a ₁, D _dis_ a ₂, D _dis_ a ₃..., D _disa _k, and by corresponding RSS vector in lastest imformation being averaged to obtain the RSS vector of reference point;

${{\mathrm{RSS}}_{\mathrm{ijp}}}^{\&prime;}=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}{\mathrm{RSS}}_{a_{k}p}/K---\left(1\right)$

In formula, p represents No. AP, RSSi _jp' representing the estimated value of the RSS value of (i, j) number p AP of reference point, the value of p is between 1～P; By as shown in the formula weighing the maturity Tr that is resumed reference point:

$\mathrm{Tr}=\frac{\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{D}_{\mathrm{dis}\_}{a}_i}{K}---\left(2\right)$

Work as Tr<Tr _thtime, think that reference point is enough ripe, can add in fingerprint image and use; Wherein, Tr _thfor with reference to thresholding, by being carried out to emulation, corresponding indoor environment obtains;

Step 2, online part

The receiver of step 2 (), user side gathers RSS vector;

The fingerprint image that step 2 (two), user use the step 1 (two) of off-line part to provide carries out fingerprint location;

Step 2 (three), utilize filtering algorithm predictive user position;

Step 2 (four), judge whether user leaves service area; Described service area refers to the region that current time fingerprint image covers;

Step 2 (five) if, upload RSS and current location, carry out fingerprint image growth operation for the step 1 (three) of off-line part; Otherwise, return to execution step two (), until user leaves the described service area of step 2 (four).

Embodiment two: present embodiment refers at the key position of building described in step 1 the region that the stream of people is larger or concentrated.

Such as being gate entrance, elevator entrance, turning, passageway, outlet or staircase, measure one region is less but fingerprint image that quality is higher at the key position (as entrance, turning etc.) of building, be referred to as " seed zone ", user can carry out fingerprint location in this region.

Embodiment three: present embodiment is to a certain degree referring to that when fingerprint image grows into maturity function Tr is less than its threshold T r described in step 1 (three) _thtime current fingerprint image.

Also the degree of closeness that can order according to uploading data point and corresponding RP judges whether enough accurately whether corresponding RP, can add in fingerprint image.

Embodiment:

Step 1, seed zone is set:

Algorithm is measured one region is less by the key position at building (as entrance, turning etc.) but fingerprint image that quality is higher, is referred to as " seed zone ", and user can carry out fingerprint location in this region.

Step 2, customer location prediction:

For motion user, due to data processing in fingerprint location process shared the regular hour, make position by calculating be user in the position in a upper moment, and real time position not.And while just leaving " seed zone " user, still can obtain by position prediction algorithm the positional information of self, gather corresponding RSS vector simultaneously.

Step 3, user data upload:

Receiver is to server uploading position information and corresponding RSS vector thereof.

Step 4, estimation reference point:

Repeating step 2 is to step 3, until in the time that the renewal sampled point in a certain region is abundant, server can calculate by them the reference point in this region.

The assessment of step 5, reference point maturity

According to the spatial distribution situation of reporting of user data and signal characteristic spatial distribution situation, judge whether this reference point has been estimated enough accurate.

Step 6, fingerprint image expand:

The reference point of enough maturations is joined in fingerprint image, and repeating step 2, to step 5, constantly expands fingerprint image and covers.

Operator carries out off-line part, and what user used is online portion, and both sides synchronously carry out, the continuous uploading position information of user and RSS, and server just constantly carries out the expansion of fingerprint image.

In implementation procedure, position prediction process can be selected particle filter, kalman filter method.The method that reference point RSS estimates can adopt KNN algorithm.

During system realizes, whole fingerprint image is saved as to the matrix of a M × N, each element of matrix has the register group that a degree of depth is K+1 to form, and each register group is made up of P+1 register, as shown in Figure 3.

As shown in FIG., front K register group storage is K RSS vector and a renewal distance B apart from this reference point of upgrading some data _dis_ a _x.Last register group internal memory be the reference point RSS vector sum maturity Tr finally calculating.

$\mathrm{Tr}=\frac{\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{D}_{\mathrm{dis}\_}{a}_i}{K}$

Its workflow is as shown in 4 figure; Tr _thfor the whether ripe threshold value of judgement reference point.

Emulation experiment and effect thereof

For the feasibility of verification algorithm in user movement environment, carry out corresponding emulation experiment herein.The virtual hall of a 30m × 50m in emulation, as shown in Figure 5.

Seed zone is 10m × 10m size, as shown in dash area in figure.Grand entrance is wide 3 meters.Experimental result, as shown in 6 figure, has shown the relation between the number of users of process in environment and area that fingerprint image covers in figure.In this experiment, the reference point quantity of off-line part collection has been reduced to 100 points from 1500, and visible this method can reduce the workload in fingerprint image process of establishing effectively.

Claims (3)

1. from the autonomous method for building up of fingerprint image of growth formula, it is characterized in that: the described autonomous method for building up of fingerprint image from growth formula is realized according to the following steps:

Step 1, off-line part

Step 1 (one), foundation " seed zone ":

Measure a basic fingerprint image at the key position of building, be referred to as " seed zone "; Wherein, the user in described seed zone can carry out fingerprint location;

Step 1 (two), determine current service district fingerprint image:

The time real growth district extending that current service district fingerprint image comprises " seed zone " and the reference point that calculates according to online certain customers end uploading data forms; Described uploading data refers to customer location coordinate and the field intensity value corresponding with this position, i.e. RSS vector; And current service district fingerprint image is offered to the user side of online part;

The detailed process of step 1 (three), fingerprint image growth:

The customer location of uploading up according to online part predicts the outcome and RSS value, and corresponding reference point RP is carried out to RSS estimation, upgrades fingerprint image, and fingerprint image growth, after fingerprint image grows into a certain degree, jumps into step 2 and redefine current service district fingerprint image;

Described customer location is uploaded information and is shown below:

A=(RSS _a1,RSS _a2,RSS _a3,…,RSS _aP,X _a,Y _a)

RSS in formula _axthe RSS value of x the AP that representative is uploaded, X _a, Y _abe respectively the coordinate information of the receipts machine estimated position reporting;

The span of x is 1～P, and P is the total number of AP in localizing environment;

It is described that corresponding reference point (RP) is carried out to the process of RSS estimation is as follows:

To a large amount of lastest imformation A that upload, try to achieve reference point locations to be estimated (Xi, Y _j) with all estimated positions (Xa, Ya) between two-dimensional distance D _dis_ a, the line item of going forward side by side;

$D_{\mathrm{dis}\_}a=\sqrt{{(X_a-X_i)}^2+{(Y_a-Y_j)}^2}$

Then system is that each reference point to be recovered is found out a nearest with it K estimated distance: D _dis_ a ₁, D _dis_ a ₂, D _dis_ a ₃..., D _dis_ a _k, and by corresponding RSS vector in lastest imformation being averaged to obtain the RSS vector of reference point;

${{\mathrm{RSS}}_{\mathrm{ijp}}}^{\&prime;}=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}{\mathrm{RSS}}_{a_{k}p}/K---\left(1\right)$

In formula, p represents No. AP, RSS _ijp' representing the estimated value of the RSS value of (i, j) number p AP of reference point, the value of p is between 1～P; By as shown in the formula weighing the maturity Tr that is resumed reference point:

$\mathrm{Tr}=\frac{\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{D}_{\mathrm{dis}\_}{a}_i}{K}---\left(2\right)$

Work as Tr<Tr _thtime, think that reference point is enough ripe, can add in fingerprint image and use; Wherein, Tr _thfor with reference to thresholding;

Step 2, online part

The receiver of step 2 (), user side gathers RSS vector;

The fingerprint image that step 2 (two), user use the step 1 (two) of off-line part to provide carries out fingerprint location;

Step 2 (three), utilize filtering algorithm predictive user position;

Step 2 (four), judge whether user leaves service area; Described service area refers to the region that current time fingerprint image covers;

Step 2 (five) if, upload RSS and current location, carry out fingerprint image growth operation for the step 1 (three) of off-line part; Otherwise, return to execution step two (), until user leaves the described service area of step 2 (four).

2. the autonomous method for building up of fingerprint image from growth formula according to claim 1, is characterized in that the key position of building described in step 1 refers to the region that the stream of people is larger or concentrated.

3. the autonomous method for building up of fingerprint image from growth formula according to claim 1 and 2, is characterized in that to a certain degree referring to that when fingerprint image grows into maturity function Tr is less than its threshold T r described in step 1 (three) _thtime current fingerprint image.

Images